The reason for the problem was  negation of signed longlong value  LONGLON
G_MIN in  Item_func_neg::int_op() - the result of this operation is not defined
(in C/C++ standard).

With this patch, LONGLONG_MIN is handled as special value, and negation is
avoided.

MDEV-4249 : when autodetecting default client charset on Windows, fallback to GetACP() whenever GetConsoleCP() returns 0 (i.e appkication does not have a console , which is the case for GUI apps, Windows services etc)

was not enabled on x86_64 machines. Enabling it. Gives up to 18 times
conversion performance improvement.

modified:
  sql/sql_string.cc

fix the declaration to use the correct type for st_handler_check_param::sort_buffer_length.
remove redundant casts.

This is a bug in the legacy code. It did not manifest itself because
it was masked by other bugs that were fixed by the patches for
mdev-4172 and mdev-4177.

Field_enum incorrectly inherited decimals() from Field_string.
Field_enum should be always integer in numeric context.

add --mysqld option to my_print_defaults
change server-postin script to use that

close (and auto-drop) temporary tables before
rolling back the last transaction in the connection.

        The bug was found by Alyssa Milburn.
        If the number of points of a geometry feature read from
        binary representation is greater than 0x10000000, then
        the (uint32) (num_points * 16) will cut the higher byte,
        which leads to various errors.
        Fixed by additional check if (num_points > max_n_points).

sql/sql_table.cc:
  Don't call allow_access_to_protected_table() if we haven't protected table against usage.
  Table is mainly protected against usage when one disables keys with alter table.

sql/sql_table.cc:
  Remove version protection from share when repair has been done.
  Without this one can't run SHOW commands on the table if it was locked until it's unlocked.
sql/table.h:
  Allow one to remove version protection with allow_access_to_protected_table()

This bug is a regression bug. The regression was introduced by
the patch for mdev-3851, that tried to weaken the condition when
a ref access with an extended key can be converted to an eq_ref
access. The patch incorrectly formed this condition. As a result,
while improving performance for some queries, the patch caused 
worse performance for another queries.

MDEV-4216 : export additional functions  mysql_get_timeout_value(),mysql_get_timeout_value_ms(),  mysql_get_socket() from shared client library. They are documented as part of async API.

Also, remove  functions mysql_close_slow_part_start() and mysql_close_slow_part_cont() from exports - they are not documented anywhere.

Removed the obsolete instructions from the MySQL 5.1 manual. Instead provide a link to https://kb.askmonty.org/en/compiling-mariadb-from-source/

The issue was that there was that SHOW commands could open the table in the store engine, even in cases
where it should not be allowed to do that (ie, the storage engines meta data for that table was under big changes).

The cases where this should not be allowed are:
- ALTER TABLE DISABLE KEYS
- ALTER TABLE ENABLE KEYS
- REPAIR TABLE
- OPTIMIZE TABLE
- DROP TABLE

This patch adds a new mode, protected_against_usage(). If this is used then the SHOW command will wait until the table
is accessable. This is implemented by re-using the already exising 'version' flag for TABLE_SHARE.
It also added functions to be used to change TABLE_SHARE->version instead of changing it directly.
	


mysql-test/r/myisam-metadata.result:
  Added test case
mysql-test/t/myisam-metadata.test:
  Added test case
sql/mysqld.cc:
  Start from refresh_version 2 as 0 and 1 are reserved.
sql/sql_admin.cc:
  Added MYSQL_OPEN_FOR_REPAIR
  Updated call to wait_while_table_is_used()
sql/sql_base.cc:
  Updated call to wait_while_table_is_used()
  - Allow one to specify how the table should be removed (for all commands except show or for all commands).
  - Don't allow one to reopen the table if one has called share->protect_against_usage()
sql/sql_base.h:
  Added TDC_RT_REMOVE_NOT_OWN_AND_MARK_NOT_USABLE, which is used to mark that no one can reopen this table, except with MYSQL_OPEN_FOR_REPAIR .
  - Added MYSQL_OPEN_FOR_REPAIR
  - Updated prototype for wait_while_table_is_used()
sql/sql_table.cc:
  Updated call to wait_while_table_is_used()
  Use MYSQL_OPEN_FOR_REPAIR for open tables that where repaired.
sql/sql_truncate.cc:
  Updated call to wait_while_table_is_used()
sql/table.cc:
  Use set_refresh_version()
sql/table.h:
  Added functions to be used to change TABLE_SHARE->version instead of changing it directly

Do not include BLOB fields into the key to access the temporary
table created for a materialized view/derived table.
BLOB components are not allowed in keys. 

  revid:georgi.kodinov@oracle.com-20120309130449-82e3bs5v3et1x0ef
  committer: Georgi Kodinov <Georgi.Kodinov@Oracle.com>
  timestamp: Fri 2012-03-09 15:04:49 +0200
  message:
    Bug #12408412: GROUP_CONCAT + ORDER BY + INPUT/OUTPUT SAME 
    USER VARIABLE = CRASH
    Moved the preparation of the variables that receive the output from 
    SELECT INTO from execution time (JOIN:execute) to compile time 
    (JOIN::prepare). This ensures that if the same variable is used in the
    SELECT part of SELECT INTO it will be properly marked as non-const
    for this query.
    Test case added.
    Used proper fast iterator.

a better fix (much smaller and without regressions) is coming from 5.1

Replace with an auto_increment primary key and another unique key didn't replicate correctly with REPLACE

MySQL Bug #12408412: GROUP_CONCAT + ORDER BY + INPUT/OUTPUT SAME USER VARIABLE = CRASH
and
MySQL Bug#14664077 SEVERE PERFORMANCE DEGRADATION IN SOME CASES WHEN USER VARIABLES ARE USED


sql/item_func.cc:
  don't use anything from Item_func_set_user_var::fix_fields()
  in Item_func_set_user_var::save_item_result()
sql/sql_class.cc:
  Call suv->save_item_result(item) *before* doing suv->fix_fields(), because
  the former evaluates the item (and caches its value), while the latter marks
  the user variable as non-const. The problem is that the item was fix_field'ed
  when the user variable was const, and it doesn't expect it to change to non-const
  in the middle of the execution.